> For the complete documentation index, see [llms.txt](https://kobotos-organization.gitbook.io/koboto-ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://kobotos-organization.gitbook.io/koboto-ai/solutions/foundational-agents.md).
# Foundational Agents
## Multi- Agent systems
**Multi-agent systems (MAS)** are composed of multiple interacting intelligent agents. These autonomous entities can sense their environment, learn models, make decisions, and take actions. Agents in an MAS can be software programs, robots, drones, sensors, humans, or a combination of these. Here are some key points about multi-agent systems:
1. **Composition**:
* MAS consists of several agents, each with specialized capabilities and goals.
* These agents collaborate to solve complex tasks that go beyond what a single agent can achieve.
2. **Key Capabilities and Benefits**:
* **Flexibility and Scalability**: MAS can adapt to changing environments by adding, removing, or modifying agents, making them highly scalable.
* **Robustness and Reliability**: Decentralization ensures continued system operation even with some failed components.
* **Self-Organization and Coordination**: Agents can self-organize based on rules for division of labor, coordinated decision-making, and conflict resolution.
***
## INFERENCE AGGREGATOR AGENT
**Inference aggregator agents** play a crucial role in combining the outputs of multiple language models (LLMs) to produce refined, high-quality responses. **Inference Aggregator Agents works same as 1inch which** provides the best token swap rates by finding optimal trading paths across multiple DEXes.
* **Process**:
* Given a prompt, they send it to multiple LLMs (which can be hosted on different platforms or systems).
* They collect the responses from these LLMs.
* Finally, they use a final LLM to combine these individual responses into an ideal answer.
**Benefits**:
* **Diverse Inputs**: Aggregating responses from different LLMs leverage their unique strengths.
* **Robustness**: Even if one LLM produces suboptimal output, the aggregation process mitigates its impact.
**Interoperability :**
* **Standardization**: Defining common protocols or APIs for communication is essential.
* **Semantic Mapping**: Creating a semantic model that maps message types (speech acts) to constraints and rational effects.
* **Feasibility Conditions**: Ensuring that sender constraints are satisfied.
**Verification**:
* **Semantic Logic (SL)**: Each message type is mapped to an SL formula defining feasibility conditions and rational effects.
* **Verification Techniques**: Techniques like formal methods, model checking, and runtime monitoring verify agent behaviour.
***
## INTENT BASED AGENTS
Unlike traditional solvers that operate on explicit commands, AI-powered solvers can understand and efficiently execute complex user intents, even when dealing with nuanced requests. Instead of merely executing transactions based on explicit commands, “intents” allow users to delegate transaction construction and execution to new actors called “solvers.” AI models equipped with natural language processing ON KOBOTO.AI can interpret these intents with a level of nuance far beyond basic instructions.
Some intent work like this , which we going to build the future of AGENTS & better UX for the fragmentation happening at layer 2 . And making MAGICSPENT++ approach more refine through agents .
Bridges & cross-chain solutions act like a duct-tape to solve this but have actually made the UX even more complicated.
For a user with 100 USDC on 3 different L2s that wants to mint an NFT for 300 USDC it’s a nightmare navigating multiple apps for this simple action. Cross-chain zaps don’t help as well since they need the user to bridge to accumulate all their balance to a single chain before they can zap — its just bridging with extra steps.
MagicSpend++, a framework to allow users to magically spend on any chain instantly, without worrying about which chains their tokens are on.
MagicSpend++ leverages the existing Account Abstraction standard and builds on Coinbase’s work on MagicSpend — an innovative approach to allow users to leverage their assets held on Coinbase Exchange & utilize them on-chain.
MagicSpend++ fundamentally enables users to have a *Chain Abstracted Balance (CAB)* instead of isolated token balances across chains. Users can use their CAB to transact instantly on any chain with a signature. No fragmentation, no bridging, no GAS, no latency. It’s magic.
Smart wallet users with MagicSpend++ get:
* A unified single balance across chains that users can spend anywhere
* Completely gas-less experience
* Instant single chain experience — NO CROSS CHAIN, NO BRIDGING, ZERO LATENCY
Usecases like magicspend with koboto agents and DeFi platforms could all users to employ AI-driven solvers through koboto to uncover intelligent ways to fulfill different intents. For example, an intent such as “I want to buy 10 ETH for no more than $3,500 per ETH” could be addressed by an AI solver that assesses a wide range of market opportunities, potentially taking on minimal risk to maximize profit and value. This would allow DeFi protocols to offer more personalized and efficient financial services, ensuring transactions are executed in alignment with users’ trading goals .
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://kobotos-organization.gitbook.io/koboto-ai/solutions/foundational-agents.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.